Lecture 12: Linear Regression with Multiple Variables Flashcards
1
Q
If alpha of gradient descent is too ? => may never find the j minimum
A
If alpha of gradient descent is too big => may never find the j minimum
2
Q
If alpha is too ?, j (cost function ) will take forever to converge
A
If alpha is too small, j (cost function ) will take forever to converge
3
Q
Feature scaling:
Make sure features are on a ? scale:
e.g. x1 = size (0-2000 m2)
=> x1 = size (m2)/2000
A
Feature scaling:
Make sure features are on a similar scale:
e.g. x1 = size (0-2000 m2)
=> x1 = size (m2)/2000
4
Q
Mean normalisation: Replace x_i with (??) with to make features have approximately ? mean (Do not apply to x0 = 1 ). e.g. x1 = size (0-2000 m2) => x1 = (size - 1000 )/ 2000
A
Mean normalisation: Replace x_i with (x_i - Mean_i) with to make features have approximately zero mean (Do not apply to x0 = 1 ). e.g. x1 = size (0-2000 m2) => x1 = (size - 1000 )/ 2000